Fatigability and muscle oxygen consumption (mVO2) during sustained voluntary isometric knee extensions are less at extended (30° knee angle; 0°, full extension) versus flexed knee angles (90°). This lower energy consumption may partially result from lower neural activation at extended knee angles. We hypothesized a smaller difference in mVO2 between extended and flexed knee angles during electrical stimulation, which guaranteed maximal activation, than during maximal voluntary contractions (MVC). In eight healthy young males, MVC extension torque was obtained at 30°, 60° and 90° knee angles. mVO2 of the rectus femoris (RF), vastus lateralis (VL) and medialis muscle was measured using near-infrared spectroscopy during tetanic (10 s) and maximal voluntary (15 s) contractions (MVC15). For electrically induced contractions, steady state mVO2 was reached at similar (P > 0.05) times after torque onset (4.6 ± 0.7 s) at all knee angles. In contrast, during MVC15 at 30° mVO2 was reached at 7.1 ± 1.1 s, significantly later compared to 60° and 90° knee angles. The knee angle dependent differences in mVO2 were not lower in electrically induced contractions (as hypothesised) but were similar as in voluntary contractions. Normalized mVO2 at 30° (percentage 90° knee angle) was 79.0 ± 9.4% (across muscles) for electrically induced and 79.5 ± 7.6% (across muscles) for voluntary contractions (P < 0.05). We conclude that the slower onset of mVO2 during voluntary effort at 30° may have been due to a lower maximal activation. However, because steady state mVO2 both during electrically induced and voluntary contractions was ~20% less at extended versus flexed knee angles, the causes for the lower mVO2 must reside within the muscle itself.